Here's a thought experiment for you: Suppose someone releases a software package under "GPL v3 or later", with all its new requirements to prevent tivoization. They do this because they like the additional guarantees their software will remain free, and they really don't want their software tivoized.
Now the FSF releases the GPL v4 that reverts the changes in GPL v3, and reads identical to GPL v2.
Theoretically, the GPL v3 was similar in spirit to the GPL v2. So, this imagined GPL v4 should also be similar in spirit to GPL v3. Now what?
Really, the "or later" clause cedes an awful lot of control to the FSF. That's been my point all along. The punch line is that if you care that much about the specific semantics of the license that covers your code, you really shouldn't put a wildcard in the license that you don't control.
Posted Mar 29, 2012 16:45 UTC (Thu) by khim (subscriber, #9252)
[Link]
The punch line is that if you care that much about the specific semantics of the license that covers your code, you really shouldn't put a wildcard in the license that you don't control.
Yup. The irony here is that it looks like FSF is the only organization which understands that.
GPL is the only popular copyleft license which gives program author the right to refuse delegation of “autoupgrade” rights to license “stewards”. Note that such liberty was given to authors by all versions of GPL, including the first one. All these other licenses were created after GPL was created - yet they cede “an awful lot of control” unconditionally.
It's funny that the very people who are bashing FSF for the “or later” clause are often advocating MPL or CC-BY-SA as an alternative despite the fact that they are obviously more problematic in this regard.